Jiaqi Mo
I'm a senior studying Computer Sciences at the University of Wisconsin–Madison, graduating in December 2026. My research focuses on efficient machine learning, from understanding model architectures to making inference and computation more efficient across software and hardware.
I currently work with Fred Sala on the design of hybrid state-space / Transformer models, and with Minjia Zhang (UIUC) on cascade inference for LLM-based kernel optimization and cloud serving for robot fleets. Before transferring to UW–Madison, I worked with Lina Wei at Hangzhou City University on Mamba-based multimodal models.
I am applying to PhD programs for Fall 2027.
News
- Paper on stateful model cascading for cost-effective kernel optimization submitted to MLSys 2027.
- Paper on synthetic languages for lightweight language model design submitted to ACL Rolling Review.
- Two papers accepted to NeurIPS 2026, including one in the Evaluations & Datasets Track.
- FORGE accepted to ASPLOS 2027.
- Started as a peer mentor for CS 540 (Introduction to Artificial Intelligence).
- MamFusion accepted to ICME 2025 as an oral presentation; follow-up work on continual audio-visual learning accepted to an ACM MM 2025 workshop.
Publications
* equal contribution
Efficient ML: Architectures and Systems
Earlier and Other Work
Teaching
Peer mentor, CS 540: Introduction to Artificial Intelligence, UW–Madison, Fall 2026.